Systematic analysis of collision-contact accident causes

Despite technological advancements and the implementation of updated regulations, collision and contact accidents at sea continue to occur due to external factors such as increasing maritime traffic density and severe meteorological conditions, as well as human errors. These types of accidents not only result in loss of life and property but also cause significant environmental disasters. Considering that such accidents typically occur as a result of a chain of interrelated causes, it is of great importance not only to identify individual causes but also to analyze the relationships between them. In this study, 79 academic articles from the Web of Science (WoS) database that focus on the causes of collision and contact accidents were examined, and the identified causes were classified within the framework of the Human Factors Analysis and Classification System (HFACS). Since the HFACS structure does not inherently reveal the relationships between causes, a further analysis was conducted to systematically present the causal relationships underlying these accidents. Among the 79 articles, 40 were identified to include inter-causal relationships. Based on these, a Bayesian Network model was developed while adhering to the HFACS-PV levels. Unlike the basic formation of the HFACS model, during the identification of causal relationships, the causes identified in the HFACS model were grouped under node names designated in accordance with the literature to enable the construction of the Bayesian structure. Through this developed model, the probabilistic relationships between causes were visualized; the strengths of the connections between nodes were indicated using color codes, thereby highlighting significant interactions. This study particularly emphasized the strong relationships between the preconditions for unsafe acts and the unsafe acts themselves. Additionally, a strong influence between training and personnel assignment on accident occurrence was observed. The connections between these two factors and ship familiarity, regional familiarity, and the company's crewing strategy were also analyzed. This research aims to contribute to the development of data-driven and holistic approaches for the prevention of collision and contact accidents and serves as a valuable resource for researchers working in this field.